WO2012142777A1

WO2012142777A1 - 8×8 integer transform method in video coding

Info

Publication number: WO2012142777A1
Application number: PCT/CN2011/073586
Authority: WO
Inventors: 徐苏珊
Original assignee: 深圳市融创天下科技股份有限公司
Priority date: 2011-04-20
Filing date: 2011-04-29
Publication date: 2012-10-26
Also published as: CN102281445A; CN102227135A

Abstract

An 8×8 integer transform method in video coding is disclosed by the present invention, which includes: receiving 8×8 image residual error data X output by a prediction module of a coder; selecting different integer transform bases [15, 15, 10, 3], [6, 7, 5, 1] to construct a transform matrix P; according to the transform matrix P, constructing a scaling matrix PF=Z^T*Z; performing forward transform on the image residual error data X according to Y=PXP^T, and adopting corresponding integer transform calculation method according to the different integer transform bases; and performing scaling processing on the 8×8 forward transformed data Y. The transform method of the present invention has excellent energy concentration performance, wherein the degree of energy concentration approaches the 8×8 Discrete Cosine Transform (DCT), thus greatly eliminating the correlation of video data in transform space. The transform method also has low calculation complexity, which can be realized by addition and shift and is easy to implement by hardware; and the transform method has high calculation precision, and thus there does not exist the problem that the forward transform does not match the inverse transform.

Description

Video coding 8x8 integer transform method

The present invention relates to the field of video coding technologies, and in particular, to a video coding 8x8 integer transform method.

Background technique

A complete video coding system consists of a series of algorithm modules such as prediction, transform, quantization, and entropy coding. Among them, intra-frame interframe prediction is compression data using spatial and temporal correlation of video data. The transform module transforms the image residual data outputted by the prediction module from the time domain to another space, so that the data in the time domain is uniformly concentrated in the transform space, and the energy of the data is mostly concentrated in the low frequency region of the space. . In the subsequent quantization and entropy coding modules, the transformed data distribution features are effectively utilized to further compress the data. Obviously, the transform module is an important part of the video coding system, and the performance of the transform directly affects the performance of the video coding system.

In the existing video coding system, the DCT (Discrete Cosine Transform) transform has been widely used, and the energy concentration of the transform is very good, which can greatly eliminate the correlation of video data in the transform space. However, this algorithm is insufficient. The DCT transformation matrix is represented by floating point. A large number of floating-point multiplications and additions are used in the calculation process, which takes up a lot of hardware resources and is computationally complex. At the same time, due to the influence of the number of operands of the processor, the floating-point operation has a truncation error, and the floating-point calculation accuracy is not high. The DCT forward-reverse transform cannot be completely matched, and the transform is irreversible. Summary of the invention

The purpose of the embodiment of the present invention is to provide a video coding 8x8 integer transform method, which aims to solve the problem that the prior art DCT transform matrix adopts floating point multiplication and addition, which occupies many hardware resources, is computationally complex, and has low calculation precision, and the DCT is positive. The inverse transformation cannot be completely matched, and the transformation is irreversible.

The method of the embodiment of the present invention is implemented as follows. A video coding 8x8 integer transform method includes:

Receiving 8x8 image residual data output by the prediction module in the encoder;

Select the integer transform base and construct the transform matrix.

The 8x8 general integer transformation matrix is: kO kO kO kO kO kO kO kO

Kl kl k3 k4 Dk4 Dk3 Uk2 Uk\

K5 k6 Uk6 Uk5 Uk5 Uk6 k6 k5

Kl Dk4 Uk\ Dk3 k3 k\ k4 Uk2

C=

kO DkO UkQ kO kO UkQ DkO kO

K3 Uk\ k4 kl Uk2 Dk4 kl Dk3

K6 Uk5 k5 Uk6 Uk6 k5 Uk5 k6

K4 Dk3 kl Uk\ k\ Uk2 k3 Dk4 In matrix C, kO, k5, k6 take values 1, 2, 1, ^{kl k3 4 respectively [} is an integer transform base; positive transform, integer transform of 8x8 image residual data , the transformation formula is yo 3⁄4P _; wherein P is the transformation matrix; is the image residual data;

The integer transform base is one of the following values: [15, 15, 10, 3], [6, 7, 5, 1].

Further, the step of "selecting an integer transform base and constructing the transform matrix P" may further include the following steps:

Construct a scaling matrix PF according to the transformation matrix P,

PFU _Z ^T ^ _Z among them,

z ^T ull\\m\\\ \l\\m2\\ l/||m3|| l/\\m4\\ l/\\m5\\ l/\\m6\\ l/\\m7\ \ l/||m8|| [; _m . is the row vector of the transformation matrix P, i takes a value of 1-8; is the vector norm of the row vector ^;

Further, the step of "positive transformation" performs integer transformation on 4x4 image residual data, and the transformation formula may further include steps:

Scale the data Y after 8x8 positive transformation,

Y UYU PF

□ is the point multiplication operation of the matrix, Y is the data after 8x8 positive transformation, PF is the scaling matrix, which is the output value after the scaling process, representing the information of the video data transformed into the frequency domain; Wherein, the basic unit of the transform in the forward transform is an 8-point one-dimensional transform of the form y = Px, where x = [xl x2 x3 x4 x5 x6 xl xSf , and the output j = [yl y2 y3 y4 y5 y6 yl J8] , l x2 x3 x4 x5 x6 χΊ J8 is the input eight-point value of the positive transform; yl yl y3 y4 _y 5 ye yl 8 is the eight-point output value of the positive transform.

When the integer transform base is [15, 15, 10, 3]:

1 1 1 1 1 1 1 1

15 15 10 3 -3 -10 -15 -15

2 1 -1 -2 -2 -1 1 2

15 -3 -15 -10 10 15 3 -15

^: Description transformation matrix P=

1 -1 -1 1 1 -1 -1 1

10 -15 3 15 -15 -3 15 -10

1 -2 2 -1 -1 2 -2 1

3 -10 15 -15 15 -15 10 -3

[0.35355 0.029907 0.22361 0.029907 0.35355 0.029907 0.22361 0.029907]

" 0.125 0.010574 0.079057 0.010574 0.125 0.010574 0.079057 0.010574

0.010574 0.00089445 0.0066875 0.00089445 0.010574 0.00089445 0.0066875 0.00089445

0.079057 0.0066875 0.05 0.0066875 0.079057 0.0066875 0.05 0.0066875

PF

0.125 0.010574 0.079057 0.010574 0.125 0.010574 0.079057 0.010574

0.079057 0.0066875 0.05 0.0066875 0.079057 0.0066875 0.05 0.0066875

"Integer transform the 8x8 image residual data, the conversion formula is y = P W; the process is as follows:

Bl=xl-x8 _; b2=x2 x3~x6; b4=x4~x5; b5=xl+x8 _; b6=x2+x7 _; b7=x3+x6 _; b8=x4+x5 _; al=b5+b8 _; a2 =b6+b7 _; a3=b5-b8 _; a4=b6_b7; yl=al+a2; y5=al~a2; y3=a3«l+a4; y7=a3—a4“l

Cl=bl«4-bl _; c2=b2«4-b2 _; c3=b3<<4_b3;c4=b4<<4_b4;

Dl=bl«3+bl«l _; d2=b2«3+b2«l; d3=b3«3+b3«l _; d4=b4«3+b4«l el=bl«l+bl _; e2= b2 «l+b2 _; e3= b3«l+b3 _; e4= b4«l+b4

Y2=cl+c2+d3 +e4; y4=cl-e2-c3_d4; y6=dl_ c2 + e3 + c4;

Y8=el - d2 + c3 - c4; where "<<" means shifting to the left, priority is higher than addition and subtraction, basic unit y ^{= Px is required}

40 additions and 14 shifts.

When the integer transform base is [6, 7, 5, 1]:

1 1 1 1 1 1 1 1

6 7 5 1 -1 -5 -7 -6

2 1 -1 -2 -2 -1 1 2

7 -1 -6 -5 5 6 1 -7

^: Description transformation matrix P=

1 -1 -1 1 1 -1 -1 1

5 -6 1 7 -7 -1 6 -5

1 -2 2 -1 -1 2 -2 1

1 -5 7 -6 6 -7 5 -1

= [0.35355 0.067116 0.22361 0.067116 0.35355 0.067116 0.22361 0.067116]

The scaling matrix is:

0.125 0.023729 0.079057 0.023729 0.125 0.023729 0.079057 0.023729

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045

0.079057 0.015008 0.05 0.015008 0.079057 0.015008 0.05 0.015008

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045

PF =

0.125 0.023729 0.079057 0.023729 0.125 0.023729 0.079057 0.023729

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045

0.079057 0.015008 0.05 0.015008 0.079057 0.015008 0.05 0.015008

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045 Image residual data Perform integer transformation, the transformation formula is r = p W. The process is as follows:

Bl=xl-x8; b2=x3-x7; b3=x3-x6; b4=x4-x5; b5=xl+x8; b6=x2+x7 b7=x3+x6; b8=x4+x5; al=b5 +b8; a2=b6+b7: a3=b5-b8; a4=b6-b7; yl=al+a2 _; y5=al-a2; y3=a3«l+a4 _; y7=a3-a4«l;

Cl=bl«2+bl«l+b4 _; c2=b2«2+b2«l-a3 ; c3=b2+b3«2+b3«l; c4 =-al+b4«2+b4«l; y2 =cl+c2+c3 ; y4=cl-c3-c4; y6=cl-c2+c4; y8=-c2+c3-c4; where "<<" means shifting to the left, priority is higher than addition and subtraction The basic unit y ^{= Px} requires a total of 32 additions and 10 shifts.

The beneficial effects of the invention:

Two sets of 8x8 integer transform matrices are proposed in the embodiments of the present invention, and a fast algorithm for transform is given. The transform method of the embodiment of the present invention has excellent energy concentration performance, and the energy concentration is close to 8x8DCT, which can greatly eliminate the transformation of video data. Spatial correlation. Low computational complexity, can be added And shift to achieve, easy to implement hardware, and high precision of operation, there is no problem of positive and negative transform mismatch.

DRAWINGS

The present invention will be further described below in conjunction with the accompanying drawings and embodiments, in which:

1 is a flowchart of a method of a preferred embodiment of a video coding 8x8 integer transform method of the present invention; FIG. 2 is a diagram of an integer transform base selection of a preferred embodiment of the video coding 8x8 integer transform method of the present invention [15, 15, 10, 3] a schematic diagram of the corresponding integer transformation calculation method;

3 is a calculation method diagram of the calculation module M1 in the schematic diagram of the integer transformation calculation method of FIG. 2; FIG. 4 is a selection of integer transformation bases for the preferred embodiment of the video coding 8x8 integer transformation method of the present invention [6, 7, 5, 1] a schematic diagram of the corresponding integer transformation calculation method;

FIG. 5 is a diagram showing a calculation method of the calculation module M2 in the schematic diagram of the integer transformation calculation method of FIG. 4.

The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. For the convenience of description, only the parts related to the embodiments of the present invention are shown. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. The invention selects different integer transform bases ([15, 15, 10, 3] or [6, 7, 5, 1]) by receiving 8x8 image residual data output by the prediction module in the encoder; for image residual data ^ Perform a positive transformation { Y = PXP ^T ), and use the corresponding integer conversion calculation method according to different integer variation bases; construct a transformation matrix Ρ, and construct a scaling matrix PF ( ^PF = z ' z ) according to the transformation matrix ;; The data Y after the 8x8 positive transformation is scaled. Embodiment 1

FIG. 1 is a video coding 8x8 integer transform method according to Embodiment 1 of the present invention, where the method includes the following steps:

S101. Receive 8×8 image residual data output by the prediction module in the encoder. S102, selecting an integer transform base, constructing a transform matrix P;

The 8x8 general integer transformation matrix is:

In matrix C, kO, k5, and k6 take values 1, 2, and 1, respectively. The definition ^{k2 k3} is constructed as a transformation matrix P for an integer transform base.

In the embodiment of the present invention, the integer transform base may select one of the following values:

[15, 15, 10, 3], [6, 7, 5, 1].

5103, constructing a scaling matrix PF according to the transformation matrix P; wherein

r = [l/||ml|| \l\\m2\\ l/||m3|| \l\\mA\\ \l\\m5\\ \l\\m6\\ \l\\ml \\ l/||m8||]; _m . is the row vector of the transformation matrix P, i takes a value of 1-8; ||^ is the vector norm of the row vector;

5104, positive transform, integer transform of 8x8 image residual data, the transformation formula is Y = PXP ^T ; where Ρ is the transformation matrix; is the image residual data;

The basic unit of transformation is an 8-point one-dimensional transformation of the form y = Px, where

x = [xl x2 x3 x4 x5 x6 xl x8 , output

j = [yl y2 y3 y4 y5 y6 yl , xl x2 x3 x4 x5 x6 xl x8 The input eight-point value of the one-dimensional forward transformation of the integer transformation; yl y3 y4 y5 y6 yl y8 is the eight-point output value of the positive transformation; According to different integer variation bases, the corresponding integer transformation calculation method is adopted.

S105, scaling the data Y after 8x8 positive transformation; Y' = Y®PF

® is the point multiplication of the matrix, Y is the 8x8 positive transformed data, PF is the scaling matrix, is the output value after the scaling process, and represents the information of the video data transformed into the frequency domain.

The transform method of the embodiment of the invention has excellent energy concentration performance, and the energy concentration is close to 8x8DCT, which can greatly eliminate the correlation of video data in the transform space; the computational complexity is low, and can be realized by addition and shifting, and is easy to implement in hardware. Moreover, the operation precision is high, and there is no problem that the forward and reverse transforms do not match. Embodiment 2

The following is a video coding 8x8 integer transform method according to Embodiment 2 of the present invention. In Embodiment 2 of the present invention, [15, 15, 10, 3] is an integer transform base of an 8x8 general integer transform matrix, and the method includes the following steps:

5201, receiving 8×8 image residual data output by the prediction module in the encoder;

5202, selecting an integer transform base, constructing a transform matrix P;

The general integer transformation matrix of 8x8 is: k0 kO kO kO kO kO kO kO

k\ k2 k3 k4 -k4 -k3 -kl -kl

K5 k6 -k6 -k5 -k5 -k6 k6 k5

K2 -k4 -kl -k3 k3 kl k4 -kl

C=

kO -kO -kO kO kO -kO -kO kO

K3 -kl k4 kl -kl -k4 kl -k3

K6 -k5 k5 -k6 -k6 k5 -k5 k6

K4 -k3 kl -kl kl -kl k3 -k4 In matrix C, k0, k5, and k6 take values 1, 2, and 1, respectively. The definition ^{k2 k3} is constructed as a transformation matrix P for an integer transform base. In the embodiment of the present invention, [15, 15, 10, 3] is selected as an integer transform base, and the transform matrix Pc is constructed. When the transform base is [15, 15, 10, 3], the transform matrix P is:

Change matrix P, construct: scaling matrix PF _;

PF = Z _z

\l\\m2\\ l/||m3|| l/\\m4\\ l/\\m5\\ l/\\m6\\ 1/ 1| m71| l/||m8||J; Ifli is the row vector of the transformation matrix P, i takes a value of 1-8; ||^ is the vector norm of the row vector;

When the transformation base of the transformation matrix P is [15, 15, 10, 3], = [0.35355 0.029907 0.22361 0.029907 0.35355 0.029907 0.22361 0.029907] The scaling matrix is:

0.125 0.010574 0.079057 0.010574 0.125 0.010574 0.079057 0.010574

0.079057 0.0066875 0.05 0.0066875 0.079057 0.0066875 0.05 0.0066875

PF

0.125 0.010574 0.079057 0.010574 0.125 0.010574 0.079057 0.010574

0.079057 0.0066875 0.05 0.0066875 0.079057 0.0066875 0.05 0.0066875

S204, positive transform, performing integer transform on 8×8 image residual data, and transforming the formula to Y=PXP ^T ; wherein Ρ is the transform matrix; and is the image residual data; The basic unit of transformation is an 8-point one-dimensional transformation of the form y ^{= Px} , where

X = [xl x2 x3 x4 x5 x6 xl xS] ^T ^ output

j = [yl y2 y3 y4 y5 y6 yl yS] ^T ^ xl xl x3 x4 x5 x6 xl x8 represents the input eight-point value of the one-dimensional forward transform of the integer transform; ^ ^{2 3 4 5} ^ is the eight-point output value of the forward transform ;

The calculation process of the integer transformation is as follows:

Bl=xl-x8 _; b2=x2-x7 _; b3=x3_x6; b4=x4_x5;

B5=xl+x8 _; b6=x2+x7 _; b7=x3+x6; b8=x4+x5;

Al=b5+b8 _; a2=b6+b7 _; a3=b5_b8; a4=b6_b7;

Yl=al+a2 _; y5=al_a2; y3=a3«l+a4; y7=a3-a4«l ;

Where bl, b2, b3, b4, b5, b6, b7, b8, al, a2, a3, a4 are intermediate variables of integer transformation;

Complete the following calculations;

Cl=bl«4-bl _; c2=b2«4-b2 _; c3=b3<<4_b3;c4=b4<<4_b4;

Dl=bl«3+bl« l _; d2=b2«3+b2«l ; d3=b3«3+b3«l _; d4=b4«3+b4«l _; el=bl«l+bl _; e2= B2«l+b2 _; e3= b3«l+b3 _; e4= b4«l+b4 _;

Y2=cl+c2+d3+e4; y4=cl-e2-c3_d4; y6=dl_c2+e3+c4;

Y8=el-d2+c3-c4;

Wherein cl, c2, c3, c4, dl, d2, d3, d4, el, e2, e3, and e4 are intermediate variables of integer transformation; Where "<<" means shifting to the left, the priority is higher than the addition and subtraction, and the basic unit y ^{= 3⁄4} requires a total of 40 additions and 14 shifts. The forward transform fast calculation method of this transform base is shown in FIG. 2, wherein the calculation module M1 is as shown in FIG. 3. Where xl x2 x3 x4 x5 x6 xl x8 represents the input eight-point value of the one-dimensional forward transform of the integer transform; yl y3 y4 y5 y6 yl y8 is the eight-point output value of the positive transform; the data processing direction is from left to right, intersecting The value of the intermediate variable of a dot is equal to the data represented by the line on the left side of the dot; the square means the data is multiplied by a coefficient; the "-" means taking a negative value; the "2" means multiplying by 2, that is, one bit to the left; "4" Represents multiplying by 4, that is, shifting two digits to the left; "8" means multiplying by 8, that is, shifting 3 digits to the left, and "16" means multiplying by 16, that is, shifting 4 digits to the left.

"inl" "in2" "in3" "in4" indicates the input data of the M1 module, which is connected to b4, b3, b2, and bl respectively; "outl" "out2" "out3" "out4" indicates the output data of the M1 module, respectively Connected to y2, y4, y6, y8.

S205, performing scaling processing on the data Y that has undergone 8×8 positive transformation;

Y' = Y®PF

The transform method of the embodiment of the invention has excellent energy concentration performance, and the energy concentration is close to 8x8DCT, which can greatly eliminate the correlation of video data in the transform space; the computational complexity is low, and can be realized by addition and shifting, and is easy to implement in hardware. Moreover, the operation precision is high, and there is no problem that the forward and reverse transforms do not match. Embodiment 3

The following is a video coding 8x8 integer transform method according to Embodiment 3 of the present invention. In Embodiment 3 of the present invention, [6, 7, 5, 1] is an integer transform base of an 8x8 general integer transform matrix, and the method includes the following steps: 5301. Receive 8×8 image residual data X output by the prediction module in the encoder.

5302, selecting an integer transform base, constructing a transform matrix P;

The general integer transformation matrix of 8x8 is: kO kO kO kO kO kO kO kO

k\ k2 k3 k4 -k4 -k3 -kl -kl

K5 k6 -k6 -k5 -k5 -k6 k6 k5

K2 -k4 -kl -k3 k3 kl k4 -kl

C=

kO -kO -kO kO kO -kO -kO kO

K3 -kl k4 kl -kl -k4 kl -k3

K6 -k5 k5 -k6 -k6 k5 -k5 k6

K4 -k3 kl -kl kl -kl k3 -k4 In matrix C, kO, k5, and k6 take values 1, 2, and 1, respectively. Define ^{k2 k3} as an integer 3 to construct a transformation matrix? .

In the embodiment of the present invention, [6, 7, 5, 1] is selected as an integer transform base, and is constructed into a transform matrix P.

When the transform base is [6, 7, 5, 1], the transform matrix P is:

1 1 1 1 1 1 1 1

6 7 5 1 -1 -5 -7 -6

2 1 -1 -2 -2 -1 1 2

7 -1 -6 -5 5 6 1 -7

1 -1 -1 1 1 -1 -1 1

5 -6 1 7 -7 -1 6 -5

1 -2 2 -1 -1 2 -2 1

1 -5 7 -6 6 -7 5 -1

S303, constructing a scaling matrix PF according to the transformation matrix P;

among them,

[1/ II ml II \l\\m2\\ l/||m3|| \l\\mA\\ \l\\m5\\ \l\\m6\\ \l\\ml\\ l/ ||m8 The row vector of the matrix P, i takes a value of 1-8; || || is the vector norm of the row vector; when the transformation base of the transformation matrix P is [6, 7, 5, 1], 0.35355 0.067116 0.22361 0.067116 0.35355 0.067116 0.22361 0.067116] The matrix is:

0.125 0.023729 0.079057 0.023729 0.125 0.023729 0.079057 0.023729

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045

0.079057 0.015008 0.05 0.015008 0.079057 0.015008 0.05 0.015008

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045

PF

0.125 0.023729 0.079057 0.023729 0.125 0.023729 0.079057 0.023729

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045

0.079057 0.015008 0.05 0.015008 0.079057 0.015008 0.05 0.015008

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045

S304, a positive transform, performing integer transform on the 8x8 image residual data ^, the transformation formula is Y = PXP ^T ; where Ρ is the transformation matrix; is the image residual data;

The basic unit of transformation is an 8-point one-dimensional transformation of the form y ^{= Px} , where

The input eight-point value of the one-dimensional positive transformation is changed; ^ y ² y ³ y ⁴ y ⁵ y ⁶ y ¹ ^ is the eight-point output value of the positive transformation;

The calculation process of the integer transformation is as follows:

Bl=xl-x8; b2=x2-x7; b3=x3-x6; b4=x4-x5;

B5=xl+x8; b6=x2+x7; b7=x3+x6; b8=x4+x5;

Al=b5+b8; a2=b6+b7; a3=b5-b8; a4=b6-b7;

Yl=al+a2 _; y5=al-a2; y3=a3«l+a4; y7=a3-a4«l;

Complete the following calculation steps:

Cl=bl«2+bl«l+b4 _; c2=b2«2+b2«l-a3 ; c3=b2+b3«2+b3«l; c4 =-al+b4«2+b4«l;

Y2=cl+c2+c3; y4=cl-c3-c4; y6=cl-c2+c4; y8=-c2+c3-c4; where cl, c2, c3, c4 are intermediate variables of integer transformation; "<<" means shifting to the left, the priority is higher than the addition and subtraction, and the basic unit y ^{= Px} requires a total of 32 additions and 10 shifts. The forward transform fast calculation method of this transform base is shown in FIG. 4, wherein the calculation module M2 is as shown in FIG. 5. Where xl x2 x3 x4 x5 x6 xl x8 represents the input four-point value of the one-dimensional forward transformation of the integer transformation; yl y3 y4 y5 y6 yl y8 is the four-point output value of the positive transformation; the data processing direction is from left to right, intersecting The value of the intermediate variable of a dot is equal to the data represented by the line on the left side of the dot; the square means the data is multiplied by a coefficient; the "-" means taking a negative value; the "2" means multiplying by 2, that is, one bit to the left; "4" Indicates multiplied by 4, that is, shifted to the left by two.

"inl" "in2" "in3" "in4" indicates the input data of the M2 module, which is connected to b4, b3, b2, and bl respectively; "outl" "out2" "out3" "out4" indicates the output data of the M2 module, respectively Connected to y2, y4, y6, y8.

S305, performing scaling processing on the data Y after 8×8 positive transformation;

Y' = Y ® PF

The transform method of the embodiment of the invention has excellent energy concentration performance, and the energy concentration is close to 8x8DCT, which can greatly eliminate the correlation of video data in the transform space; the computational complexity is low, and the addition can be performed by adding It is realized by shifting, easy to implement in hardware, and has high precision of operation, and there is no problem that the forward and reverse transforms do not match. It will be understood by those skilled in the art that all or part of the steps of the foregoing embodiments may be implemented by a program instruction related hardware, and the program may be stored in a computer readable storage medium. The storage medium may be a ROM, a RAM, a magnetic disk, an optical disk, or the like.

The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. Within the scope.

Claims

Rights request

1, a video encoding 8x8 integer transform method, including

Select the integer transform base and construct the transform matrix P.

The general integer transformation matrix of 8x8 is: k0 kO kO kO kO kO kO kO

k\ k2 k3 k4 -k4 -k3 -kl -kl

K5 k6 -k6 -k5 -k5 -k6 k6 k5

K2 -k4 -kl -k3 k3 kl k4 -kl

kO -kO -kO kO kO -kO -kO kO

K3 -kl k4 kl -kl -k4 kl -k3

K6 -k5 k5 -k6 -k6 k5 -k5 k6

K4 -k3 kl -kl kl -kl k3 -k4 In matrix C, k0, k5, k6 take values 1, 2, 1, - ^{k k2 k3} for integer transform positive transform, integer transform for 8x8 image residual data , the transformation formula is Γ = Ρ Ρ where P is the transformation matrix; is the image residual data;

It is characterized by:

The integer transform base is one of the following values:

[15, 15, 10, 3], [6, 7, 5, 1].

2. The video coding 8x8 integer transform method according to claim 1, wherein: i "selecting an integer transform base, constructing a transform matrix P" may further comprise the steps of:

According to the transformation matrix P, a scaling matrix PF is constructed, where ml|| \l\\m2\\ l/||m3|| \l\\mA\\ \l\\m5\\ \l\\m6\\ \l\\ml\\ l/||m8||]; To transform the row vector of the matrix P, i takes a value of 1-8; ||^ is the vector norm of the row vector.

The video coding 8x8 integer transform method according to claim 2, wherein the step of "selecting an integer transform base, constructing the transform matrix P" may further comprise the steps of: performing data 8 after 8x8 forward transform Zoom processing, Y' = Y®PF

4. The video coding 8x8 integer transform method according to claim 1, wherein the basic unit of the transform in the forward transform is an 8-point one-dimensional transform of the form y = Px, where x = [xl x2 x3 X4 x5 x6 xl 8f , output y = [yl yl y3 y4 y5 y6 yl ySf , xl x2 x3 x4 x5 x6 xl J8 is the input eight-point value of the forward transform; yl yl y3 y4 y5 y6 yl y8 is The eight-point output value of the positive transform.

The video coding 8x8 integer transform method according to claim 2, wherein when the integer transform base is [15, 15, 10, 3]:

1 1 1 1 1 1 1 1

15 15 10 3 -3 -10 -15 -15

2 1 -1 -2 -2 -1 1 2

15 -3 -15 -10 10 15 3 -15

^: Description transformation matrix P=

1 -1 -1 1 1 -1 -1 1

10 -15 3 15 -15 -3 15 -10

1 -2 2 -1 -1 2 -2 1

3 -10 15 -15 15 -15 10 -3

= [0.35355 0.029907 0.22361 0.029907 0.35355 0.029907 0.22361 0.029907]

0.125 0.010574 0.079057 0.010574 0.125 0.010574 0.079057 0.010574

0.079057 0.0066875 0.05 0.0066875 0.079057 0.0066875 0.05 0.0066875

PF =

0.125 0.010574 0.079057 0.010574 0.125 0.010574 0.079057 0.010574

0.079057 0.0066875 0.05 0.0066875 0.079057 0.0066875 0.05 0.0066875

6. The video coding 8x8 integer transform method according to claim 1, wherein when the integer transform base value is [15, 15, 10, 3], "integer for 8x8 image residual data. Transform, transform formula for r = PW calculation process is as follows:

Bl= =xl-x8; b2= =x2~x7; b3= =x3~x6; b4= =x4~x5; b5= =xl+x8; b6= =x2+x7; b7= =x3+x6; b8 ==x4+x5; al==b5+b8 _; a2= =b6+b7 _; a3= =b5-b8 _; a4= =b6-b7 _; yl= =al+a2 _; y5= =al~a2; y3= =a3«l+a4 _; y7=a3-a4«l _; again; into the following drive [step:

Cl=bl«4-bl _; c2=b2«4-b2 _; c3=b3«4-b3 _; c4=b4<<4_b4; dl=bl«3+bl«l _; d2=b2«3+b2«l ; d3=b3«3+b3«l _; d4=b4«3+b4«l el=bl«l+bl _; e2= b2«l+b2 _; e3= b3«l+b3 _; e4= b4«l+ B4 y2=cl+c2+d3 +e4; y4=cl-e2-c3_d4; y6=dl_ c2 + e3 + c4; y8=el - d2 + c3 - c4;

Where "<<" means shifting to the left, the priority is higher than the addition and subtraction.

7. The video coding 8x8 integer transform method according to claim 2, wherein: When the integer transform base is [6, 7, 5, 1]:

1 1 1 1 1 1 1 1

6 7 5 1 -1 -5 -7 -6

2 1 -1 -2 -2 -1 1 2

7 -1 -6 -5 5 6 1 -7

^: Description transformation matrix P=

1 -1 -1 1 1 -1 -1 1

5 -6 1 7 -7 -1 6 -5

1 -2 2 -1 -1 2 -2 1

1 -5 7 -6 6 -7 5 -1 = [0.35355 0.067116 0.22361 0.067116 0.35355 0.067116 0.22361 0.067116] The scaling matrix is:

0.125 0.023729 0.079057 0.023729 0.125 0.023729 0.079057 0.023729

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045

0.079057 0.015008 0.05 0.015008 0.079057 0.015008 0.05 0.015008

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045

PF

0.125 0.023729 0.079057 0.023729 0.125 0.023729 0.079057 0.023729

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045

0.079057 0.015008 0.05 0.015008 0.079057 0.015008 0.05 0.015008

0.023729 0.0045045 0.015008 0.0045045 0.023729 0.0045045 0.015008 0.0045045

8. The video coding 8x8 integer transform method according to claim 1, wherein when the integer transform base value is [6, 7, 5, 1], "integer change is performed on 8x8 image residual data. The transformation formula is y = PXP ^T "The calculation process is as follows: bl=xl-x8; b2= =x2-x7; b3=x3-x6; b4=x4-x5;

B5=xl+x8; 6 =x2+x7; b7=x3+x6; b8=x4+x5;

Al=b5+b8; a2==b6+b7; a3=b5-b8; _a 4=b6-b7;

Yl=al+a2; y5==al-a2; y3=a3«l+a4 _; y7=a3-a4«l; Complete the following calculation steps:

y y y y;;;ccc3c2l24lc3c6ccc4l24∞cc3c2++++-----====

Cabb4l424l++««-

Cbl=